×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Fusion of sensor data, knowledge sources and algorithms for extraction and classification of topographic objects
Author
Baltsavias, Emmanuel P.

International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999
wiridwurf JjJ |(none]
1
1
ChanneH Mean
J UlkiiTTTiTS
C
3
).0 31.9 63.8 ^5.6 127.5 159.4 191.3 223.1 255.0
3 87.8
Channels Mean
[27.4■ 35.3] StdAbw.: 1.95
2
>.0 21.6 ■j'38.3 54.9 71.5 88.1 104.3 121.4 138.0
4 32.5
Channel4Mean
^[25.2 - 36.0] StdAbw.: 2.47
4
i.O 22.1 f 41.3 60.4 79.5 93.6 117.8 136.9 156.0
1 30.9
Channels Mean
[107.0-134.0] StdAbw: 6.20
/' l\
2
ie.O 53.8 81.5 109.3 1^.0 164.8 192.5 220.3 248.0
5 138.4
Channels Mean
/-■“"Tt--- [100.9 - 122.5] StdAbw.: 6.28
X akijj K
1
2.0 44.9 67.8 90.6 fji.5 136.4 159.3 182.1 205.0
0 113.1
ChanneH Std.Dev.
r, - J98.4] StdAblAi.: 13.52
l If! 1 1 till ,
C
1
.0 16.0 32.0 48.0 64.0 80.0 96^ 112.0 128.0
6 99.2
Channel!' Std.Dev.
Jr" TTT---. [15.2 - 39.7] StdAbvu.: 7.55
,Xh iii iiilj n " . 1
0.0 10.1 20.3^30.4 40.5 50.6 60.8 70.9 S1.0
25.2
The fuzzy logic functions are shown with the following information:
Channel 1: Texture image using the first moment (Haralick et al., 1973)
Channel 2: Panchromatic orthophoto (only standard deviation is used)
Channel 3: SPOT4 band 1
Channel 4: SPOT4 band 2
Channel 5: SPOT4 band 3
Channel 6: SPOT4 band 4
Fig. 2. The same forest road as in Fig. 1 at the outmost left side. The fuzzy logic decision rules for wind-damaged areas
(‘windwurf) on the right. The SPOT4 pixels are regarded as spatial objects and the panchromatic information is also used
to classify ‘windwurf. This area is represented in white. The spectral characteristics are very similar to several other
landcover classes in the area. The spatial structure of many dead tree stems has an impact on the texture analysis using the
first moment filter operation. Each double black line in the bar graph shows the value of a test object. The fuzzy logic
functions can then be calculated and manually altered to increase class discrimination possibilities. Fuzzy logic functions
can also be set without any sampled area. This makes the set of rules transferable to other image data as well. The spatial
rule "number of sub-objects >25" (see hierarchy features in Fig. 1) is not shown here, but used in the analysis.
204